On Fri, Jan 20, 2006 at 09:51:40PM -0500, Derrick J Brashear wrote: > Do you have your fixes somewhere?
On Fri, Jan 20, 2006 at 06:53:34PM -0800, Kevin Elliott wrote: > Any chance you're willing to share your new bug free code? :) On Sat, Jan 21, 2006 at 09:53:38AM +0530, hemendra wrote: > even i would like to see the code sample..if you are willing > i m also a part of this community I wouldn't have mentioned it if I wasn't going to share. If you follow the link I provided to the CVS repository for the source code, you'll see the code and the diffs. See http://cvs.ofb.net/geocoder/. My C skills aren't the greatest, but hopefully this will be helpful. The bug had to do with a delta adjustment to the coordinates in the data compression that was supposed to be a signed byte, but it was being fed values in the range -255 to +255, which wasn't right. This was causing results to be off in an X or Y direction by a few city blocks sometimes. The other issue wasn't a bug, per se, but the TIGER data structure had changed a bit in 2004, and some adjustment in the record parsing had to be made. The changes in the TIGER 2005 First Edition from the 2004 Second Edition won't affect this code, but it does bring up a point: TIGER structure changes occasionally and any robust code designed to use it should either be version-aware, or be able to handle different versions flexibly. - Bill Thoen _______________________________________________ Geowanking mailing list [email protected] http://lists.burri.to/mailman/listinfo/geowanking
